"Anshin EURJPY"

Outline] We have developed a highly reliable EA that is low stress and can be used with confidence from a small budget. It checks in advance when the trade logic stops working and changes the judgment criteria. It is a type of EA that does not increase assets all at once, but accumulates profits steadily every day.　Feature.1] There is always one entry per day, whether to sell or buy. Entry is at midnight and close is in the morning.　<メリット>　(1) Even an ordinary office worker can check entries late at night Japan time and check the status in the morning. (2) Only one position, either buy or sell, is required to start with a small budget. When you have enough funds, you can increase the number of lots. (3) EA can be stopped and rested on days when the market is moving rapidly, such as when important economic indicators are announced. (4) Losses are small, so there is no stress caused by unrealized losses. (5) Stress is low because you know when to enter and when to close once a day. Feature 2] We have improved the reliability of EAs through pseudo-forward testing.　Many EAs perform well in back testing, but tend to perform poorly in forward testing. Therefore, we decided to optimize and create an EA using only the first 9 years of historical data out of 10 years of data, and then pseudo-forward test the EA with the remaining historical data of the most recent year.　The method was to use the EA with the best back-test results and the best pseudo-forward test results. We used 187 technical indicators. ( <メリット>1) To confirm that the optimization of parameters is not curve fitting. (2) It is highly reliable and easy to set parameters because it is simulated and analyzed with many technical indicators. (3) Periodic pseudo-forward tests can be conducted to check if the EA's logic is no longer valid, and optimal parameter settings can be made. Since this EA only requires a buy or sell decision once a day, we narrowed down the decision criteria to 187 and conducted a simulation using EXCEL VBA.　The main ones are: (1) Moving averages (SMA, EMA, SMMA, LWT, HULL, Original] (2) Bolivan 2σ, 3σ (3) CCI (4) RSI (5) Equilibrium table, with different periods and time frames, 187 in total. Currency pair: EURJPY Time axis: M30 Historical data: (2005.10.7~2015.10.6) 10 years Spread: 15 Of the 10 years, we used the first 9 years of data and back-tested it by making a decision to buy or sell at 9 years, 5 years, 4 years, 3 years, 2 years and 1 year for each criteria (187 pieces). I tested it back to back. The PF (Profit Factor) and Profit (Profit:US$) were obtained. PF7 and Profit7 are the results of the pseudo-forward test for the most recent year. In all periods, the PF exceeded 1 and its value was large, and the fluctuation of the PF value was small. We decided to use the criterion ID:139 for the first EA release v1. The 10-year back-test results of the first EA release above are as follows: (1) Profit: 12,918 (US$) (2) PF: 1.30 (3) Number of trades: 2,593 (buy: 1827, sell: 766) (4) Win rate: 53.5% (5) Maximum drawdown: 958 (US$) US$) [Parameter Description and Default Values] - Lots = 0.1 Number of lots - TakeProfit = 140 Limit value (pips) - StopLoss = 100 Loss Cut value (pips) - EntryHalfhour = 0 Entry time (Japan time, in 30-minute units)　(e.g. Enter 0 to set the entry time to 0:00, 15 to set the entry time to 1:30, etc.) CloseHalfhour = 90 Close time (Japanese time, in units of 30 minutes)　　(ex. Enter 90 for 9:00, 75 for 7:30, etc.) AutoTimeSet ：true When backtesting, set this value to false and enter a value for TimeDifferece. Backtests are not supported for historical data, which differ from Japan time in summer and winter, such as GMT+3 for summer time and GMT+2 for winter time. Please make sure to set it to "true" for normal demo and real trades. TimeDifference ： 6 The difference between Japan time (GMT+9) and the server time recorded in the historical data.　MaxSpread = 4.0 If the spread is greater than this value, the trade will not be executed. Slippage = 3.0 Slippage value ・ MagicNumber = 51115102 Magic Number Please be sure to set a unique value to avoid mixing with other orders. Do not change TakeProfit, StopLoss, EntryHalfhour, and CloseHalfhour significantly, and use the default values as much as possible. The logic (decision criteria) will surely reach a point where it no longer works, so we will recalculate the above pseudo forward test every month, and if it is necessary to change the criteria, we will respond by upgrading the EA.